identifAI Launches REVELIO V Plus and REVELIO Face II: A New Frontier in Deepfake Detection

identifAI today announced the simultaneous release of two of its most advanced AI-generated content detection models to date: REVELIO V Plus, a next-generation system for detecting AI-generated images and video, and REVELIO Face II, a highly specialised model for identifying AI-generated face manipulations. Together, they represent a significant step forward in the fight against synthetic media.

The Challenge We Are Solving

Synthetic media is no longer a niche concern. Advances in generative AI have made it possible for virtually anyone to produce convincing deepfakes — fabricated images, manipulated videos, and artificially generated faces that are increasingly indistinguishable from authentic content.

The consequences are far-reaching. From disinformation campaigns and identity fraud to reputational damage and legal disputes, the misuse of AI-generated content poses a genuine and growing threat to individuals, organisations, and public discourse alike.

Research consistently shows how difficult the problem has become: humans fail to correctly identify high-quality deepfakes 76% of the time. Reliable, automated detection is no longer optional — it is essential.

Two REVELIO Models, One Mission

REVELIO V Plus and REVELIO Face II address two distinct but complementary dimensions of the deepfake detection challenge.

• REVELIO V Plus is designed for the broad detection of AI-generated images and video. It analyses content at scale, identifying synthetic material produced by a wide range of generative models — including the latest and most widely used tools available today.
• REVELIO Face II addresses a more targeted problem: the detection of AI-generated face manipulations. Whether a face has been entirely synthesised or subtly altered, REVELIO Face II is built to identify these manipulations with a level of precision and speed that no previous identifAI model has achieved.

‍

What Is New in REVELIO V Plus

REVELIO V Plus delivers meaningful improvements across every dimension of performance compared to its predecessor.

The model has been retrained to recognise 15 new AI generators, including Grok Imagine, Nanobanana 2, GPT Image 2, and Seedance 2 — ensuring that detection capabilities keep pace with the rapidly expanding landscape of generative tools. It also delivers fast inference times and higher detection accuracy than any previous version of REVELIO, making it well suited to high-volume, real-time workflows.

The results speak for themselves. REVELIO V Plus achieves an overall accuracy of 99.8%, with an Artificial F1 Score of 99.9%, Precision of 100.0%, and Recall of 99.9%. Measured across both public and private datasets, it represents a 7.4% performance gain over REVELIO V.

What Is New in REVELIO Face II

REVELIO Face II has been built from the ground up for the specific challenges of face manipulation detection.

The model is fast and more accurate than any face detection system identifAI has previously released. It covers a broader range of AI generators and manipulation techniques, reflecting how quickly methods for synthesising and altering faces continue to evolve.

A particular area of innovation is the model's optimisation for minimum face size. Detecting AI manipulation in very small faces — those appearing at low resolution or occupying only a small portion of a frame — has historically been one of the hardest problems in this domain. REVELIO Face II has been specifically engineered to address this, significantly extending detection capability beyond what earlier models could achieve.

REVELIO Face II achieves an overall accuracy of 95.5%, representing a 13.9% performance gain over its predecessor across both public and private datasets.

Why This Matters

The release of REVELIO V Plus and REVELIO Face II reflects identifAI's commitment to developing detection technology that keeps pace with the generative AI landscape — not reactively, but proactively.

As generative models become faster, more accessible, and more capable of producing convincing synthetic content, the tools used to detect them must evolve in parallel. These two models are designed not only to address the threats that exist today, but to provide a foundation robust enough to meet those that will emerge tomorrow.

Both models are available now on the identifAI platform.

For further information or to arrange a demonstration for your organisation, contact sales@identifai.net or visit identifai.net.